An empirical item characteristic curve shows the probability of a correct response as a function of the student's total test score. These curves can be estimated from large-scale pretest data. They enable test developers to select items that discriminate well in the score region where decisions are made. A similar set of curves can be used to predict the equating relationship between two forms of the test. These curves are based on the common-item score instead of the total score. They make it possible to estimate raw-to-scale score conversions before the test is administered. (Author)